alignmenter 0.1.1__tar.gz → 0.2.0__tar.gz

{alignmenter-0.1.1/src/alignmenter.egg-info → alignmenter-0.2.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alignmenter
-Version: 0.1.1
+Version: 0.2.0
 Summary: Persona-aligned evaluation toolkit for auditing conversational AI authenticity, safety, and stability.
 Author: Alignmenter
 License:                                  Apache License
@@ -209,8 +209,8 @@ Project-URL: Homepage, https://alignmenter.com
 Project-URL: Repository, https://github.com/justinGrosvenor/alignmenter
 Project-URL: Documentation, https://github.com/justinGrosvenor/alignmenter
 Project-URL: Bug Tracker, https://github.com/justinGrosvenor/alignmenter/issues
-Keywords: llm,evaluation,alignment,persona,safety
-Classifier: Development Status :: 3 - Alpha
+Keywords: llm,evaluation,alignment,persona,safety,llm-judge
+Classifier: Development Status :: 4 - Beta
 Classifier: Intended Audience :: Developers
 Classifier: Intended Audience :: Information Technology
 Classifier: License :: OSI Approved :: Apache Software License
@@ -219,35 +219,49 @@ Classifier: Programming Language :: Python
 Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: typer[all]
-Requires-Dist: pydantic
-Requires-Dist: pydantic-settings>=2.0
-Requires-Dist: openai>=1.12.0
-Requires-Dist: anthropic>=0.18.0
-Requires-Dist: scikit-learn>=1.3.0
-Requires-Dist: numpy>=1.24.0
-Requires-Dist: pyyaml>=6.0
-Requires-Dist: sentence-transformers>=2.2.2
-Requires-Dist: torch
-Requires-Dist: tiktoken>=0.7.0
-Requires-Dist: requests>=2.32.0
+Requires-Dist: typer[all]<1,>=0.12
+Requires-Dist: pydantic<3,>=2
+Requires-Dist: pydantic-settings<3,>=2
+Requires-Dist: openai<3,>=1.40
+Requires-Dist: anthropic<1,>=0.40
+Requires-Dist: pyyaml<7,>=6
+Requires-Dist: tiktoken<1,>=0.7
+Requires-Dist: requests<3,>=2.32
+Provides-Extra: ml
+Requires-Dist: torch<3,>=2; extra == "ml"
+Requires-Dist: sentence-transformers<6,>=3; extra == "ml"
+Requires-Dist: transformers<6,>=4.40; extra == "ml"
+Provides-Extra: calibrate
+Requires-Dist: scikit-learn<2,>=1.3; extra == "calibrate"
+Requires-Dist: numpy<3,>=1.24; extra == "calibrate"
+Provides-Extra: safety
+Requires-Dist: alignmenter[ml]; extra == "safety"
+Provides-Extra: all
+Requires-Dist: alignmenter[calibrate,ml]; extra == "all"
 Provides-Extra: dev
-Requires-Dist: pytest; extra == "dev"
+Requires-Dist: alignmenter[calibrate,ml]; extra == "dev"
+Requires-Dist: pytest>=8; extra == "dev"
 Requires-Dist: pytest-cov; extra == "dev"
-Requires-Dist: ruff; extra == "dev"
+Requires-Dist: ruff>=0.6; extra == "dev"
 Requires-Dist: build; extra == "dev"
 Requires-Dist: twine; extra == "dev"
-Provides-Extra: safety
-Requires-Dist: transformers>=4.30.0; extra == "safety"
 Dynamic: license-file
 
 <p align="center">
   <img src="https://alignmenter-branding.s3.us-west-2.amazonaws.com/alignmenter-banner.png" alt="Alignmenter" width="800">
 </p>
 
+<p align="center">
+  <a href="https://pypi.org/project/alignmenter/"><img src="https://badge.fury.io/py/alignmenter.svg" alt="PyPI version"></a>
+  <a href="https://pepy.tech/project/alignmenter"><img src="https://pepy.tech/badge/alignmenter" alt="Downloads"></a>
+  <a href="https://github.com/justinGrosvenor/alignmenter/blob/main/LICENSE"><img src="https://img.shields.io/badge/License-Apache_2.0-blue.svg" alt="License"></a>
+</p>
+
 <p align="center">
   <strong>Persona-aligned evaluation for conversational AI</strong>
 </p>
@@ -285,19 +299,37 @@ cd alignmenter
 python -m venv env
 source env/bin/activate  # On Windows: env\Scripts\activate
 
-# Install with dev + safety extras
-pip install -e ./alignmenter[dev,safety]
+# Install with all optional extras for development
+pip install -e "./alignmenter[dev]"
 ```
 
 ### Install from PyPI
 
+The core install is lightweight — no `torch`, no `scikit-learn`. It runs the
+default scoring path (hashed embeddings + keyword safety, optionally blended
+with an LLM judge) out of the box:
+
 ```bash
-pip install "alignmenter[safety]"
+pip install alignmenter
 alignmenter init
+alignmenter run --config configs/run.yaml
+```
+
+Add optional extras only when you need them:
+
+| Extra | Adds | Use it for |
+|-------|------|------------|
+| `alignmenter[ml]` | torch, sentence-transformers, transformers | Local embeddings (`--embedding sentence-transformer:...`) and the offline safety classifier |
+| `alignmenter[calibrate]` | scikit-learn, numpy | The persona calibration pipeline (`calibrate*` commands) |
+| `alignmenter[all]` | both of the above | Everything |
+
+```bash
+# Example: better local embeddings + offline safety classifier
+pip install "alignmenter[ml]"
 alignmenter run --config configs/run.yaml --embedding sentence-transformer:all-MiniLM-L6-v2
 ```
 
-> **Note**: The `safety` extra includes `transformers` for the offline safety classifier (ProtectAI/distilled-safety-roberta). Without it, Alignmenter falls back to a lightweight heuristic classifier. See [docs/offline_safety.md](https://github.com/justinGrosvenor/alignmenter/blob/main/docs/offline_safety.md) for details.
+> **Offline safety classifier**: `[ml]` includes `transformers` for the offline classifier (ProtectAI/distilled-safety-roberta). Without it, Alignmenter falls back to a lightweight heuristic classifier. See [docs/offline_safety.md](https://github.com/justinGrosvenor/alignmenter/blob/main/docs/offline_safety.md) for details.
 
 ### Run Your First Evaluation
 
@@ -324,10 +356,6 @@ alignmenter dataset sanitize datasets/demo_conversations.jsonl --out datasets/de
 alignmenter run --config configs/run.yaml --generate-transcripts
 ```
 
-## Case Studies
-
-- [Wendy's Twitter Voice](../docs/case-studies/wendys-twitter.md) – end-to-end calibration example using the included case-study assets. *(Available when running from the source repo; not included in the PyPI wheel.)*
-
 **Output:**
 ```
 Loading dataset: 60 turns across 10 sessions
@@ -443,6 +471,14 @@ Quick links:
 - **[LLM Judges](https://docs.alignmenter.com/guides/llm-judges/)** - Qualitative analysis
 - **[Contributing](https://docs.alignmenter.com/contributing/)** - How to contribute
 
+---
+
+## Case Studies
+
+- **[Wendy's Twitter Voice](../docs/case-studies/wendys-twitter.md)** - End-to-end calibration example using the included case-study assets. *(Available when running from the source repo; not included in the PyPI wheel.)*
+
+---
+
 ## Usage Examples
 
 ### Evaluate Multiple Models
@@ -541,32 +577,43 @@ traits:
 
 ## API Usage
 
+`Runner` coordinates transcript preparation, scoring, and report generation.
+It takes a `RunConfig` plus a list of scorers, and `execute()` returns the
+path to the timestamped report directory (JSON + HTML are written for you).
+
 ```python
-from alignmenter.runner import Runner
-from alignmenter.config import RunConfig
-
-# Load configuration
-config = RunConfig.from_yaml("configs/run/my_eval.yaml")
-
-# Execute evaluation
-runner = Runner(config)
-results = runner.execute()
-
-# Access scores
-print(f"Authenticity: {results['scores']['authenticity']['mean']:.3f}")
-print(f"Safety: {results['scores']['safety']['fused_judge']:.3f}")
-print(f"Stability: {results['scores']['stability']['session_variance']:.3f}")
-
-# Generate reports
-from alignmenter.reporting import HTMLReporter, JSONReporter
-
-html_reporter = HTMLReporter()
-html_reporter.write(
-    run_dir=results["run_dir"],
-    summary=results["summary"],
-    scores=results["scores"],
-    sessions=results["sessions"],
+import json
+from pathlib import Path
+
+from alignmenter.runner import RunConfig, Runner
+from alignmenter.scorers.authenticity import AuthenticityScorer
+from alignmenter.scorers.safety import SafetyScorer
+from alignmenter.scorers.stability import StabilityScorer
+
+config = RunConfig(
+    model="openai:gpt-4o-mini",
+    dataset_path=Path("datasets/demo_conversations.jsonl"),
+    persona_path=Path("configs/persona/default.yaml"),
 )
+
+# Pass a judge to AuthenticityScorer/SafetyScorer to blend LLM judgment in;
+# omit it (as here) for a fully offline, deterministic run.
+scorers = [
+    AuthenticityScorer(persona_path=config.persona_path, embedding="hashed"),
+    SafetyScorer(keyword_path=Path("configs/safety_keywords.yaml")),
+    StabilityScorer(embedding="hashed"),
+]
+
+# generate_transcripts=False reuses recorded transcripts (no provider calls).
+runner = Runner(config, scorers, generate_transcripts=False)
+run_dir = runner.execute()  # -> Path to reports/<timestamp>_<run_id>/
+
+results = json.loads((run_dir / "results.json").read_text())
+primary = results["scores"]["primary"]
+auth = primary["authenticity"]
+print(f"Authenticity: {auth['mean']:.3f}  (basis: {auth['basis']})")
+print(f"Safety:       {primary['safety']['score']:.3f}")
+print(f"Stability:    {primary['stability']['stability']:.3f}")
 ```
 
 ## CI/CD Integration
@@ -649,13 +696,25 @@ make report-last
 
 ## Roadmap
 
-- [ ] Web dashboard for team collaboration
-- [ ] Hosted evaluation pipelines (CI/CD SaaS)
-- [ ] Multi-language support (non-English personas)
-- [ ] Offline safety model (no API required)
-- [ ] Cost/latency tracking in scorers
-- [ ] Competitive landscape analysis
-- [ ] Metric cards and ablation studies
+### Completed ✅
+- Three-dimensional scoring (authenticity, safety, stability)
+- Multi-provider support (OpenAI, Anthropic, local models)
+- HTML report cards with interactive charts
+- Offline safety classifier (distilled-safety-roberta)
+- LLM judges for qualitative analysis
+- Budget guardrails and cost tracking
+- PII sanitization tools
+- Calibration workflow and diagnostics
+
+### In Progress 🚧
+- Multi-language support (non-English personas)
+- Batch processing optimizations
+- Additional embedding providers
+
+### Future Considerations 💭
+- Synthetic test case generation
+- Custom metric plugins
+- Advanced trait models (neural networks)
 
 ## Contributing
 
@@ -679,14 +738,15 @@ If you use Alignmenter in research, please cite:
 @software{alignmenter2024,
   title={Alignmenter: A Framework for Persona-Aligned Conversational AI Evaluation},
   author={Alignmenter Contributors},
-  year={2024},
-  url={https://github.com/justinGrosvenor/alignmenter}
+  year={2025},
+  url={https://github.com/justinGrosvenor/alignmenter},
+  license={Apache-2.0}
 }
 ```
 
 ## Support
 
-- **Documentation**: [docs/](https://github.com/justinGrosvenor/alignmenter/tree/main/docs)
+- **Documentation**: [docs.alignmenter.com](https://docs.alignmenter.com)
 - **Issues**: [GitHub Issues](https://github.com/justinGrosvenor/alignmenter/issues)
 - **Discussions**: [GitHub Discussions](https://github.com/justinGrosvenor/alignmenter/discussions)
 

{alignmenter-0.1.1 → alignmenter-0.2.0}/README.md

@@ -2,6 +2,12 @@
   <img src="https://alignmenter-branding.s3.us-west-2.amazonaws.com/alignmenter-banner.png" alt="Alignmenter" width="800">
 </p>
 
+<p align="center">
+  <a href="https://pypi.org/project/alignmenter/"><img src="https://badge.fury.io/py/alignmenter.svg" alt="PyPI version"></a>
+  <a href="https://pepy.tech/project/alignmenter"><img src="https://pepy.tech/badge/alignmenter" alt="Downloads"></a>
+  <a href="https://github.com/justinGrosvenor/alignmenter/blob/main/LICENSE"><img src="https://img.shields.io/badge/License-Apache_2.0-blue.svg" alt="License"></a>
+</p>
+
 <p align="center">
   <strong>Persona-aligned evaluation for conversational AI</strong>
 </p>
@@ -39,19 +45,37 @@ cd alignmenter
 python -m venv env
 source env/bin/activate  # On Windows: env\Scripts\activate
 
-# Install with dev + safety extras
-pip install -e ./alignmenter[dev,safety]
+# Install with all optional extras for development
+pip install -e "./alignmenter[dev]"
 ```
 
 ### Install from PyPI
 
+The core install is lightweight — no `torch`, no `scikit-learn`. It runs the
+default scoring path (hashed embeddings + keyword safety, optionally blended
+with an LLM judge) out of the box:
+
 ```bash
-pip install "alignmenter[safety]"
+pip install alignmenter
 alignmenter init
+alignmenter run --config configs/run.yaml
+```
+
+Add optional extras only when you need them:
+
+| Extra | Adds | Use it for |
+|-------|------|------------|
+| `alignmenter[ml]` | torch, sentence-transformers, transformers | Local embeddings (`--embedding sentence-transformer:...`) and the offline safety classifier |
+| `alignmenter[calibrate]` | scikit-learn, numpy | The persona calibration pipeline (`calibrate*` commands) |
+| `alignmenter[all]` | both of the above | Everything |
+
+```bash
+# Example: better local embeddings + offline safety classifier
+pip install "alignmenter[ml]"
 alignmenter run --config configs/run.yaml --embedding sentence-transformer:all-MiniLM-L6-v2
 ```
 
-> **Note**: The `safety` extra includes `transformers` for the offline safety classifier (ProtectAI/distilled-safety-roberta). Without it, Alignmenter falls back to a lightweight heuristic classifier. See [docs/offline_safety.md](https://github.com/justinGrosvenor/alignmenter/blob/main/docs/offline_safety.md) for details.
+> **Offline safety classifier**: `[ml]` includes `transformers` for the offline classifier (ProtectAI/distilled-safety-roberta). Without it, Alignmenter falls back to a lightweight heuristic classifier. See [docs/offline_safety.md](https://github.com/justinGrosvenor/alignmenter/blob/main/docs/offline_safety.md) for details.
 
 ### Run Your First Evaluation
 
@@ -78,10 +102,6 @@ alignmenter dataset sanitize datasets/demo_conversations.jsonl --out datasets/de
 alignmenter run --config configs/run.yaml --generate-transcripts
 ```
 
-## Case Studies
-
-- [Wendy's Twitter Voice](../docs/case-studies/wendys-twitter.md) – end-to-end calibration example using the included case-study assets. *(Available when running from the source repo; not included in the PyPI wheel.)*
-
 **Output:**
 ```
 Loading dataset: 60 turns across 10 sessions
@@ -197,6 +217,14 @@ Quick links:
 - **[LLM Judges](https://docs.alignmenter.com/guides/llm-judges/)** - Qualitative analysis
 - **[Contributing](https://docs.alignmenter.com/contributing/)** - How to contribute
 
+---
+
+## Case Studies
+
+- **[Wendy's Twitter Voice](../docs/case-studies/wendys-twitter.md)** - End-to-end calibration example using the included case-study assets. *(Available when running from the source repo; not included in the PyPI wheel.)*
+
+---
+
 ## Usage Examples
 
 ### Evaluate Multiple Models
@@ -295,32 +323,43 @@ traits:
 
 ## API Usage
 
+`Runner` coordinates transcript preparation, scoring, and report generation.
+It takes a `RunConfig` plus a list of scorers, and `execute()` returns the
+path to the timestamped report directory (JSON + HTML are written for you).
+
 ```python
-from alignmenter.runner import Runner
-from alignmenter.config import RunConfig
-
-# Load configuration
-config = RunConfig.from_yaml("configs/run/my_eval.yaml")
-
-# Execute evaluation
-runner = Runner(config)
-results = runner.execute()
-
-# Access scores
-print(f"Authenticity: {results['scores']['authenticity']['mean']:.3f}")
-print(f"Safety: {results['scores']['safety']['fused_judge']:.3f}")
-print(f"Stability: {results['scores']['stability']['session_variance']:.3f}")
-
-# Generate reports
-from alignmenter.reporting import HTMLReporter, JSONReporter
-
-html_reporter = HTMLReporter()
-html_reporter.write(
-    run_dir=results["run_dir"],
-    summary=results["summary"],
-    scores=results["scores"],
-    sessions=results["sessions"],
+import json
+from pathlib import Path
+
+from alignmenter.runner import RunConfig, Runner
+from alignmenter.scorers.authenticity import AuthenticityScorer
+from alignmenter.scorers.safety import SafetyScorer
+from alignmenter.scorers.stability import StabilityScorer
+
+config = RunConfig(
+    model="openai:gpt-4o-mini",
+    dataset_path=Path("datasets/demo_conversations.jsonl"),
+    persona_path=Path("configs/persona/default.yaml"),
 )
+
+# Pass a judge to AuthenticityScorer/SafetyScorer to blend LLM judgment in;
+# omit it (as here) for a fully offline, deterministic run.
+scorers = [
+    AuthenticityScorer(persona_path=config.persona_path, embedding="hashed"),
+    SafetyScorer(keyword_path=Path("configs/safety_keywords.yaml")),
+    StabilityScorer(embedding="hashed"),
+]
+
+# generate_transcripts=False reuses recorded transcripts (no provider calls).
+runner = Runner(config, scorers, generate_transcripts=False)
+run_dir = runner.execute()  # -> Path to reports/<timestamp>_<run_id>/
+
+results = json.loads((run_dir / "results.json").read_text())
+primary = results["scores"]["primary"]
+auth = primary["authenticity"]
+print(f"Authenticity: {auth['mean']:.3f}  (basis: {auth['basis']})")
+print(f"Safety:       {primary['safety']['score']:.3f}")
+print(f"Stability:    {primary['stability']['stability']:.3f}")
 ```
 
 ## CI/CD Integration
@@ -403,13 +442,25 @@ make report-last
 
 ## Roadmap
 
-- [ ] Web dashboard for team collaboration
-- [ ] Hosted evaluation pipelines (CI/CD SaaS)
-- [ ] Multi-language support (non-English personas)
-- [ ] Offline safety model (no API required)
-- [ ] Cost/latency tracking in scorers
-- [ ] Competitive landscape analysis
-- [ ] Metric cards and ablation studies
+### Completed ✅
+- Three-dimensional scoring (authenticity, safety, stability)
+- Multi-provider support (OpenAI, Anthropic, local models)
+- HTML report cards with interactive charts
+- Offline safety classifier (distilled-safety-roberta)
+- LLM judges for qualitative analysis
+- Budget guardrails and cost tracking
+- PII sanitization tools
+- Calibration workflow and diagnostics
+
+### In Progress 🚧
+- Multi-language support (non-English personas)
+- Batch processing optimizations
+- Additional embedding providers
+
+### Future Considerations 💭
+- Synthetic test case generation
+- Custom metric plugins
+- Advanced trait models (neural networks)
 
 ## Contributing
 
@@ -433,14 +484,15 @@ If you use Alignmenter in research, please cite:
 @software{alignmenter2024,
   title={Alignmenter: A Framework for Persona-Aligned Conversational AI Evaluation},
   author={Alignmenter Contributors},
-  year={2024},
-  url={https://github.com/justinGrosvenor/alignmenter}
+  year={2025},
+  url={https://github.com/justinGrosvenor/alignmenter},
+  license={Apache-2.0}
 }
 ```
 
 ## Support
 
-- **Documentation**: [docs/](https://github.com/justinGrosvenor/alignmenter/tree/main/docs)
+- **Documentation**: [docs.alignmenter.com](https://docs.alignmenter.com)
 - **Issues**: [GitHub Issues](https://github.com/justinGrosvenor/alignmenter/issues)
 - **Discussions**: [GitHub Discussions](https://github.com/justinGrosvenor/alignmenter/discussions)
 

alignmenter-0.2.0/pyproject.toml

@@ -0,0 +1,110 @@
+[build-system]
+requires = ["setuptools>=68", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "alignmenter"
+version = "0.2.0"
+description = "Persona-aligned evaluation toolkit for auditing conversational AI authenticity, safety, and stability."
+authors = [{name = "Alignmenter"}]
+readme = {file = "README.md", content-type = "text/markdown"}
+requires-python = ">=3.10"
+license = {file = "LICENSE"}
+keywords = ["llm", "evaluation", "alignment", "persona", "safety", "llm-judge"]
+classifiers = [
+  "Development Status :: 4 - Beta",
+  "Intended Audience :: Developers",
+  "Intended Audience :: Information Technology",
+  "License :: OSI Approved :: Apache Software License",
+  "Topic :: Scientific/Engineering :: Artificial Intelligence",
+  "Programming Language :: Python",
+  "Programming Language :: Python :: 3",
+  "Programming Language :: Python :: 3.10",
+  "Programming Language :: Python :: 3.11",
+  "Programming Language :: Python :: 3.12",
+  "Programming Language :: Python :: 3.13",
+]
+# Core install is lightweight: no torch, no scikit-learn. The default scoring
+# path (hashed embeddings + LLM judge + keyword safety) runs on these alone.
+# Heavy local-model dependencies live in the optional [ml] and [calibrate] extras.
+dependencies = [
+  "typer[all]>=0.12,<1",
+  "pydantic>=2,<3",
+  "pydantic-settings>=2,<3",
+  "openai>=1.40,<3",
+  "anthropic>=0.40,<1",
+  "pyyaml>=6,<7",
+  "tiktoken>=0.7,<1",
+  "requests>=2.32,<3",
+]
+
+[project.optional-dependencies]
+# Local embeddings (sentence-transformers) + offline safety classifier (transformers).
+# This is the only extra that pulls in torch.
+ml = [
+  "torch>=2,<3",
+  "sentence-transformers>=3,<6",
+  "transformers>=4.40,<6",
+]
+# Persona calibration / trait-model training and diagnostics.
+calibrate = [
+  "scikit-learn>=1.3,<2",
+  "numpy>=1.24,<3",
+]
+# Back-compat alias: the offline safety classifier ships in [ml].
+safety = ["alignmenter[ml]"]
+# Everything runtime-optional in one shot.
+all = ["alignmenter[ml,calibrate]"]
+dev = [
+  "alignmenter[ml,calibrate]",
+  "pytest>=8",
+  "pytest-cov",
+  "ruff>=0.6",
+  "build",
+  "twine",
+]
+
+[project.urls]
+Homepage = "https://alignmenter.com"
+Repository = "https://github.com/justinGrosvenor/alignmenter"
+Documentation = "https://github.com/justinGrosvenor/alignmenter"
+"Bug Tracker" = "https://github.com/justinGrosvenor/alignmenter/issues"
+
+[project.scripts]
+alignmenter = "alignmenter.cli:app"
+
+[tool.setuptools]
+include-package-data = true
+
+[tool.setuptools.package-dir]
+"" = "src"
+
+[tool.setuptools.packages.find]
+where = ["src"]
+
+[tool.setuptools.package-data]
+alignmenter = [
+  "data/configs/**/*.yaml",
+  "data/configs/**/*.txt",
+  "data/datasets/*.jsonl",
+]
+
+[tool.ruff]
+line-length = 100
+target-version = "py310"
+src = ["src", "tests"]
+
+[tool.ruff.lint]
+select = ["E", "F", "I", "UP", "B"]
+ignore = ["E501", "B008"]
+
+[tool.ruff.lint.per-file-ignores]
+# Typer Exit/BadParameter are control-flow re-raises; chaining a traceback adds noise.
+"src/alignmenter/cli.py" = ["B904"]
+"src/alignmenter/scripts/run_openai_demo.py" = ["B904"]
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]
+filterwarnings = [
+  "error::DeprecationWarning:alignmenter.*",
+]

{alignmenter-0.1.1 → alignmenter-0.2.0}/src/alignmenter/calibration/analyze.py

@@ -5,11 +5,10 @@ from __future__ import annotations
 import json
 from collections import defaultdict
 from pathlib import Path
-from typing import Optional
 
-from alignmenter.scorers.authenticity import AuthenticityScorer
-from alignmenter.providers.judges import load_judge_provider
 from alignmenter.judges.authenticity_judge import AuthenticityJudge
+from alignmenter.providers.judges import load_judge_provider
+from alignmenter.scorers.authenticity import AuthenticityScorer
 
 
 def analyze_scenario_performance(
@@ -17,10 +16,10 @@ def analyze_scenario_performance(
     persona_path: Path,
     output_path: Path,
     *,
-    embedding_provider: Optional[str] = None,
+    embedding_provider: str | None = None,
     judge_provider: str,
     samples_per_scenario: int = 3,
-    judge_budget: Optional[int] = None,
+    judge_budget: int | None = None,
 ) -> dict:
     """
     Analyze performance across different scenario types.
@@ -39,7 +38,7 @@ def analyze_scenario_performance(
     """
     # Load dataset
     sessions = []
-    with open(dataset_path, "r") as f:
+    with open(dataset_path) as f:
         for line in f:
             if line.strip():
                 session = json.loads(line)

{alignmenter-0.1.1 → alignmenter-0.2.0}/src/alignmenter/calibration/bounds.py

@@ -5,12 +5,25 @@ from __future__ import annotations
 import json
 import math
 from pathlib import Path
-from typing import Optional
 
-import numpy as np
+try:
+    import numpy as np
+except ModuleNotFoundError as _exc:  # pragma: no cover - exercised without [calibrate]
+    _CALIBRATE_IMPORT_ERROR: ModuleNotFoundError | None = _exc
+    np = None  # type: ignore[assignment]
+else:
+    _CALIBRATE_IMPORT_ERROR = None
 
 from alignmenter.providers.embeddings import load_embedding_provider
 from alignmenter.utils import load_yaml
+from alignmenter.utils.optional import missing_dependency
+
+
+def _require_calibrate() -> None:
+    if _CALIBRATE_IMPORT_ERROR is not None:
+        raise missing_dependency(
+            "Persona calibration", "calibrate", "scikit-learn, numpy"
+        ) from _CALIBRATE_IMPORT_ERROR
 
 
 def estimate_bounds(
@@ -18,7 +31,7 @@ def estimate_bounds(
     persona_path: Path,
     output_path: Path,
     *,
-    embedding_provider: Optional[str] = None,
+    embedding_provider: str | None = None,
     percentile_low: float = 5.0,
     percentile_high: float = 95.0,
 ) -> dict:
@@ -38,9 +51,10 @@ def estimate_bounds(
     Returns:
         Bounds report with statistics
     """
+    _require_calibrate()
     # Load labeled data
     labeled_data = []
-    with open(labeled_path, "r") as f:
+    with open(labeled_path) as f:
         for line in f:
             if line.strip():
                 labeled_data.append(json.loads(line))
@@ -133,7 +147,7 @@ def estimate_bounds(
     with open(output_path, "w") as f:
         json.dump(report, f, indent=2)
 
-    print(f"\n✓ Bounds estimation complete")
+    print("\n✓ Bounds estimation complete")
     print(f"  Style similarity min: {style_sim_min:.4f}")
     print(f"  Style similarity max: {style_sim_max:.4f}")
     print(f"  Mean: {report['style_sim_mean']:.4f}")